96k is better than 44.1k sample rate. why?

Reggie said:
BTW: I would worry a lot more about using 96K myself if people in the mainstream could listen to 96k CDs in their stereos.
Click to expand...

I'm all for it, and that'll be the day I start tracking everything at 96K.
 
Reggie said:
This is some of the stuff that I've never been able to quite get my head around. Say you have a source that contains some 20k information. But as with many natural sounds it isn't perfect sine waves. Would a 44.1k sample rate be unable to convey the little imperfections since it is only sampling enough times to basically encode some sort of 20kHz sound, thereby causing the AD or DA convertors to sort of guess what shape the wave really is?
Aaaaand my head exploded.


BTW: I would worry a lot more about using 96K myself if people in the mainstream could listen to 96k CDs in their stereos.
Click to expand...

you have to understand that 20kHz is very very high in the frequency spectrum of music. This is mainly just going to be upper harmonics of instruments and not the fundamental note. Again, not many instruments really play much higher than the C8 note on the piano (around 4000Hz). The upper harmonics of a fundamental note is what gives each instrument their own timbre. And there are nuances up there that strengthen the sound of the instrument.
The middle C on piano is where a lot of instruments reside, and this is around 261Hz. This frequency's 76th harmonic is where it nears 20,000. Not a huge deal if we are losing some of that. Now yes, around C8...the 5th harmonic is nearer to 20,000 and the instrument may start to sound different than when we listen to it in the room. But again, human hearing (perfect hearing) is between 20-20kHz...even though most people don't hear that high. This range is one of the reasons 44.1kHz was decided upon...and then we got a safety net of 48kHz to use as well if you so choose (most people do).

look into anti-aliasing to help with your converter question
 
But 10kHz will have 4.41 samples taken of each of it's waveforms. Now look at the first waveform we drew. In that drawing we took 6 samples of the waveform and got an amplitude reading saying 0,2,2,0,2,2. imagine how inaccurate 4.41 samples are of a complex waveform. That is why digital high frequencies sound harsh!! The industry has constantly denied this factor and even gone to the extent of saying the hear can't distinguish between a square wave and a sine wave above 7kHz. Pigs Bum.

At a sampling rate of 96kHz you get 9.6 samples of a 10kHz wave and believe me, you can hear it.

In an article by Rupert Neve, I read recently, he said that we should aim for 24bit resolution and 192kHz sampling rate if we want to equal the quality of high quality analogue recording. We will get there. DVD is already up to 24 bit 96kHz sampling so we are on the way. But if your 16bit, 44.1kHz CD sounds bright, consider what makes it bright and you will see that it's a false bright created by the high frequencies sounding like square waves!!
Click to expand...


it looks like that guy has no idea what he's talking about, frequency is to sampling rate as apples are to CARS!!
 
guitarfreak12 said:
it looks like that guy has no idea what he's talking about, frequency is to sampling rate as apples are to CARS!!
Click to expand...

well, balancing sample rate and frequency IS important. It helps avoid aliasing.
 
96k or 44.1k you ask??
1)you are going to mixdown to 44.1 anyway.
2)DAW won't be able to handle as many tracks
3)computer's processor has to work harder.
4)you cant hear it.

i dont care what anyone says, when it's all said and done, you would never be able to hear the difference between a session recorded in 44.1, 48, 88, 96, 192, whatever. its going to be 44.1 anyways......
 
VSProductions said:
96k or 44.1k you ask??
1)you are going to mixdown to 44.1 anyway.
2)DAW won't be able to handle as many tracks
3)computer's processor has to work harder.
4)you cant hear it.

i dont care what anyone says, when it's all said and done, you would never be able to hear the difference between a session recorded in 44.1, 48, 88, 96, 192, whatever. its going to be 44.1 anyways......
Click to expand...

1 - Yes you are! And retaining the HIGHER QUALITY sounding tracks up until your mix has been mastered is ALWAYS important!

2 - Nope, it won't! 96K is NOT for the faint of heart with DAW's! ;)

3 - Yes, it does. But, you more or less already covered that with 2 eh?

4 - Maybe YOU can't hear it, but, there are brilliant engineers who CAN! Nuf said. MANY engineers that are MUCH better than you and I can hear the difference. I can hear the difference. I don't know why you can't. Maybe a monitoring problem in your studio or something.
 
tossing in my 2 cents

I use 192khz at 24 bit for record restoration....when filtering out the noise (snap crackle and pop)....more samples the better i get better results doing it that way.

also I can make a DVD audio disk as well closer to the record its self. :D
were 44.1khz at 16 bit is chunky in comparison. :rolleyes:

the weird class out there audiophiles :p which want audio DVDS :cool:

I record at 96khz at 24 bit ....then mix down to to 44.1khz at 16bit.

but I can understand why a lot would record at 44.1 at 16 bit as well.
It also makes a lot of sense in that you dont have to down convert the files.
 
bennychico11 said:
Not in your situation. Especially not with a synth. The highest note on the synth wouldn't even come close to the highest reproducable frequency at 48kHz sampling rate. The highest note on a typical piano is C8 which is 4,186Hz. No where near the Nyquist Frequency of 24kHz.
Click to expand...

That is it's fundamental, which says nothing about it's harmonics.

I can hear a tiny difference between 44.1khz and 96khz/88.2khz. Could just be my converters though. You also have to consider the fact that during the D/A conversion the uppermost frequencies turn into perfect sine waves because of the filter. While put through the right filter this might not sound "bad", but it's definitely not accurate.
 
It all depends on the quality of the filters and converters.

Nonetheless, one should not attempt to record to 16 bit, because the cumulative loss of fidelity due the quantization errors is horrible. 24 bit is the key for recording.

About sample rate, if you record at 44.1, even the best converter and filter will exhibit a treble roll-off at about 18 kHz (maybe less), because no filter can be made ideal. When you think of it, this is why CDplayers do oversample the 44.1 signals about 8 times to be able to design a much simpler filter. Having that said, 48 is the minimum I would go for tracking and mixing. Preferably, I would choose 88.2, since the downsampling to 44.1 is much simpler than with 96 or 48 since it only chops one sample out of two.

About the sound quality, I can hear the differences between 88.2, 48 and 41 pretty well. Past 88.2, I can't hear a significant improvement at all. I doubt anyone who can hear improvements over 88.2 (or 96). At that high of a sampling rate, you simply don't miss much out of the music and the filtering is much easier to do.

To sum this up, the higher you can sample is always the better, but improvements heard for sampling rates over 88.2 are mostly due to poor converters and filters to begin with. I personnaly use 48/24 because it's a good compromise on the strain you put on your DAW and the sound qualty. If I had a 10k RPM SATA drive with 300GB capacity and one of those AMD mega processors, I would use 88.2/24 no questions asked.

As a side note, sound quality improvement between 24 and 32 bit is very hard to discern. Some might say it "opens up the top end", but I say the improvement I hear doesn't warrant the extra strain on my DAW. Again, if I had "THE DAW" I would probably use 32 bits.
 
Ford Van said:
4 - Maybe YOU can't hear it, but, there are brilliant engineers who CAN! Nuf said. MANY engineers that are MUCH better than you and I can hear the difference. I can hear the difference. I don't know why you can't. Maybe a monitoring problem in your studio or something.
Click to expand...

Sure there is a difference between 44.1 and higher sample rates, but it's not at 7 kHz and 10 kHz like the rather ill-informed link argued. I would challenge anyone to here a difference at 44.1 and higher rates with bandwidth-limited material below 10kHz.

Dan Lavry has often stated that 50-60kHz was more than adequate to address the concerns with 44.1 and 48 causing attenuation of high frequencies. My DAW is happy working at 64kHz, but my converters don't like it :( For them it's 48 or 88.1, nothing in between. I believe we should start demanding that converters function at 64kHz, but I am strange.

Anyway, I decided to put Lavry to the test. Here are some FFT plots of white noise at 96kHz, vs. the same file downsampled and then upsampled. Note as a control, I generated a 44.1kHz white noise file and upsampled it, to eliminate the loss on downsampling. That showed that downsampling (at least my downsampling routine) performs a little worse than theoretically perfect, but I think the SRC is more relevant anyway.

Why? Because a number of plugs run at higher internal sample rates. Thus your audio might suffer several SRCs of which you are unaware, and the effect is cumulative.

So here is 44.1kHz:
 
Personally, I use 48kHz. I would like to use 64kHz or higher, say 88.2kHz, but that causes me a couple of problems--not processing power, I don't run that many tracks, and I have 2 UAD-1s anyway :) But I'd drop from 10 input channels to 6, and from 8 outputs to 4. So I couldn't monitor a surround 96kHz mix.

Also I only have the HD24 in my mobile rig, not the XR, so that is limited to 48kHz.

So therefore I keep life simple by recording at 48kHz, because while not perfect, it really isn't bad at all. I can hear 48kHz vs. 44.1kHz, but I can't distinguish between 48 and higher rates.

Here's 48kHz:
 
These graphs show exactly what I was saying before. A slight attenuation around 18kHz. Nice to know I'm not a fool!
 
Cool ideas. But I'm curious what you used for SRC? It might be neat to compare something like Barbabatch vs a good hardware SRC. just to see if things can be any better. I'm all for a 60kHz sample rate, but I don't see it happening since so much 96k is in place.



As a side note, sound quality improvement between 24 and 32 bit is very hard to discern.
Click to expand...

Where do you get these wonderful 32-bit convertors? Or are you referring to mixing down several 24bit files to a 32-bit mix?


Preferably, I would choose 88.2, since the downsampling to 44.1 is much simpler than with 96 or 48 since it only chops one sample out of two.
Click to expand...

Actually, most experts (not me) will tell you this is a myth. The math is much different than simply dividing a number in half. Otherwise sample rate conversion wouldn't be nearly as complicated a deal and there wouldn't be all these different brands and software programs that do it differently.
 
neat stuff. I might have to try my next project at 48kHz. I don't think I can hear much of those frequencies though. My ears aren't what they used to be. I've played with too many loud drummers.
 
Reggie said:
Where do you get these wonderful 32-bit convertors? Or are you referring to mixing down several 24bit files to a 32-bit mix?

Actually, most experts (not me) will tell you this is a myth. The math is much different than simply dividing a number in half. Otherwise sample rate conversion wouldn't be nearly as complicated a deal and there wouldn't be all these different brands and software programs that do it differently.
Click to expand...
1) I'm talking about using 32bit float option instead of 24 bits.

2) A myth? Sample rate conversion is complicated whenever there isn't a common integer by which you can divide the samples. In that case, you have to actually choose which sample you keep, while having twice the sample frequency allows you to keep one sampe out of two.
 
Reggie said:
Cool ideas. But I'm curious what you used for SRC? It might be neat to compare something like Barbabatch vs a good hardware SRC. just to see if things can be any better. I'm all for a 60kHz sample rate, but I don't see it happening since so much 96k is in place.
Click to expand...

This is the Resampler 192kHz bundled with Wavelab. It's the better of Wavelab's two SRCs, but obviously it still isn't a great SRC. By comparison, the UAD-1 Precision EQ upsamples to 192kHz, and a before-and-after graph shows no change at all (other than the EQ change selected). Makes me wish they split out their SRC into a plug!

But the point of those graphs was really sort of a worst-case or maybe below average-case scenario. That sort of attenuation could be happening with a converter or an SRC.

Ultimately, I'm trying to address the question of why people feel on listening tests that 96 is better than 44.1. Sometimes people say that you can't here the difference on a single track, but you can on a mix. There must be a theoretical reason for that, if it is true, but I think this is one possible explanation.

I've done a test on A/D converters before that show a similar phenomenon, that was 44.1 vs. 48 on an Alesis AI3. The attenuation at 44.1 was nowhere near as bad, but it was there, and it was audible too. That was the test that convinced me to use 48.

Since then I've upgraded my converters, I might have time on Sunday to test my new converters (RME ADI-8), which also do 88.2 and 96, so that would be interesting to see how those perform.
 
You must log in or register to reply here.
Back
Top